Batch Correction While Batch Numbers and Labels Match

Question

[ghiasirad.milad]

Hello all,

I have combined two subsets of two different GSE datasets (One as Controls and One as Cases) and have made a new data frame. I see a complete separation between the two batches. So, I thought I should do the batch correction. But the batch indices and labels completely match. The resulting batch corrected dataset against those labels (which were the batches too) returned no DEGs. Would someone explain why this has happened and how I can get rid of the batch effect for this specific case, please? Thanks in advance.

Answer 1

[Papyrus]

If all cases belong to a batch, and all controls belong to a different batch, it is impossible to do a batch correction, because you can't know if the differences between the 2 groups are due to the batch, or to their experimental conditions (i.e. the conditions are "totally confounded"). And vice-versa: the same thing applies for any difference you may find between the 2 groups and think to be biological.

When you tried to correct the differences between the 2 batches, you cancelled-out the differences between the 2 groups (because they are the same thing), that's why you did not find any differences.

[ATpoint]

+1

That analysis makes no sense, data must come from the same study or all DEGs youfind could be due to technical differences, no way to tell.